Give away medical masks when you place an order. learn more

Selecting a Personal Area Network Link for Wearables and the Internet of Things

The real deployment of personal area networks (PANs) has not yet taken hold. We are on the cusp of a new generation of wearable computers, sensors, and peripherals that will intertwine us with machines at a new level.

Traditionally, a PAN would be associated with a wireless audio link such as a Bluetooth connection to a wireless headset. While this is a useful, local machine-to-machine (M2M) link for an individual, it falls far short of the real potential that a low-power RF near-field technology PAN can provide.

This article will examine the choices we have for generating and transceiving data throughout our “personal electromagnetic bubble.” It examines the use of low power levels and signal types for a variety of applications ranging from wearable or “deeply embedded” sensors on the simple side, to more complex high-definition video and image processing for real-time 3D gesture recognition.

We will examine present-day chip-level solutions for standards such as IRDA, Wireless USB, Bluetooth, Z-Wave, ZigBee, and Wi-Fi, paying close attention from a bandwidth perspective to determine what is the real, usable throughput we can expect to see. We will also look at which standards are better for different functions. All parts, datasheets, tutorials, and development tools referenced here can be found on the Hotenda website.

Not everything is RF

Wireless links typically invoke the notion of radio, but not every wireless link is RF based. Some line-of-sight, short-hop, low-bandwidth communications can be IR based instead. Take, for instance, a two-part force feedback glove for remote control of equipment or medical procedures. Here, an IRDA module like the ROHM RPM973-H11E2A would do a good job (Figure 1). This transceiver is ultra-thin and self-contained, and can provide up to 4 Mbits/s as an optical link that will not be interfered with by ambient RF noise from any source. It is also ruggedly constructed for harsh conditions.

Figure 1: Do not discount tough and rugged IR as a useful line-of-sight link for modest-bandwidth data communications. Several well-engineered low-cost transceivers are available for engineers to specify.

While optical technologies have their place, by far the most widely used communications technology for emerging PAN applications will be RF. It is interesting to note that for low-speed links at very short distances, lower-cost narrow-band AM, FM, ASK, FSK, carrier on/carrier off, and PSK types of RF can be used. A computer mouse works just fine at 1,200-bit/s data rates.

The Murata TR3000 supports data rates up to 115.2 Kbaud using a 433.92 MHz carrier and ASK, or OOK modulation. Operating at 2.7 to 3.7 V, it draws a modest 3.8 mA during receive and can burst out transmissions at 7.5 mA draw. A nice feature is that power usage can be scaled back for very-short-distance links, extending battery life (Figure 2).

Figure 2: Narrow-band transmissions can use very low amounts of power for relatively low data rates. Noise sources and crowded interoperable environments pose problems, however.

While the power limitations of narrow-band AM and FM could work, there are too many possible sources of interference and these types of links typically will not have arbitration, collision detection, collision avoidance, and automatic retransmission when errors occur. Here is where digital radios will shine.

Several digital standards are vying for the coveted, potentially very-high-volume PAN market, including consistently interoperable standards like Bluetooth, USB, ZigBee, Wi-Fi, or Z-Wave, to name a few.

Wireless USB offers some promise, with several IC level devices ready to step in. Consider the Cypress CYRF6936-40LTXC direct-sequenced spread-spectrum wireless 2.4 GHz USB transceiver-only part. With data rates up to 1 Mbit/s, the 1.8 to 3.6 V unit uses a 4 MHz SPI port for setup and control. It comes in an exposed-pad 40-pin part that is a bit larger than a narrow-band solution. Its 34 mA transmit (and 21.2 mA receive) currents are significantly higher as well. However, many applications will spend more time asleep than awake; and when awake, communications bursts can go on for a long time on small batteries, especially if they are rechargeable.

A similar part with an embedded controller is the Cypress CYRF89235-40LTXC, which provides an up-to-24 MHz Harvard architecture M8C RISC processor on-chip, as well as emulation ports (Figure 3). On-chip 32 K Flash can house stack and user code for some apps. The 2 K RAM can be expanded in programmed I/O through the 8-bit ports or through I²C or SPI interfaces, which are also included.

Figure 3: The system-on-a-chip approach allows the embedded micro to completely run the protocol stack while providing an embedded environment to either house your application-specific code or create your own custom interface.

Beyond audio

Bluetooth audio will most likely remain dominant for headsets and for on-person audio links, even though it typically uses more power than is needed. Bluetooth devices, for the most part, play well together, even in crowded environments. The tethering process lets transceivers be simple look-and-lock types without having to maintain multiple sockets and complex protocol stacks.

Bluetooth Low Energy, on the other hand, is well suited for nonaudio applications like sensors, actuators, and PANs. Similar to other standards, integrated solutions are ready to step up to the plate. One Bluetooth LE solution of note comes from CSR with its TCSR1010A05-IQQM-R single-chip Bluetooth LE system on a chip (SoC) transceiver (Figure 4). Part of CSR’s µEnergy Bluetooth Low Energy Platform, it also contains an embedded microcontroller, in this case a 16-bit RISC processor that runs the BT LE stack, radio, interrupts, and external interfaces.

Figure 4: Embedded micros not only can contain the digital radio peripheral functionality, but also provide other connectivity and peripheral interfaces, including mixed signal.

It should be noted that these parts have a bit more resources available; Flash is 64 Kbytes and RAM is 64 Kbytes. In addition, these parts also contain a 10-bit A/D, 12 programmable I/Os, SPI, I²C, UART, PWM, and a debug SPI port. As with virtually every radio transceiver being developed today, they also have energy management features and can use 32 kHz real-time clock crystals for extended-sleep power savings.

Another competitor in this space is STMicroelectronics with its Bluetooth LE wireless network processor, the BLUENRGQTR. Also complying with the Bluetooth v4.0 spec as a 1 Mbit/s compliant master or slave, it can use a 32 kHz clock or oscillator for energy reduction or run on its higher native frequency for process-intensive crunching, in this case up to 32 MHz.

It is based on an ARM Cortex-M0 processor (Figure 5), with a usable memory of 64 K program Flash and 12 K SRAM. It also has SPI, I²C, UART, serial program and debug, as well as AES hardware. STMicroelectronics sees this as a potential PAN field peripheral controller, especially for health and fitness applications. The company also offers a Product Training Module for Health and Fitness applications of Bluetooth LE.

Figure 5: Not only are 8- and 16-bit cores finding their way into PAN applications, this 32-bit Cortex-M0 can operate radio links and have plenty of processing power for your code.

Like several other suppliers, STMicroelectronics supports the stack and offers a development environment to help you get up to speed quickly. In this case, the supplier’s STEVAL-IDB002V1 is a useful demo and evaluation board for BlueNRG low-energy network processors.

Other possibilities

Other wireless players that want in on the newly exploding PAN market have some hurdles to overcome. One such case is ZigBee, which is a popular standard supported by dozens, or device and module manufacturers for home and building automation.

Unlike Bluetooth, ZigBee does not have native support on Smartphones, tablets, and laptops. This may present an obstacle. ZigBee also requires a rather sophisticated stack, meaning node costs could be higher. On the other hand, ZigBee offers the advantage of being part of a large mesh with architected arbitration and identification.

Wi-Fi also has a certain appeal, especially with the push toward the Internet of Things. It offers cloud-based connectivity and its chips and modules are available as ready-to-go certified solutions. While it is supported in native Smartphones, tablets, and notebooks, Wi-Fi uses a lot of power. Flexible control can scale it back for PAN applications, but this still would not be a viable answer if, after every time it goes into low power mode, it needs to re-establish connection when waking up; discovery mode takes time and power.

There are other potential solutions. Z-Wave, ANT +, IOHomecontrol, W6LoPAN, and RF4CE are among the application- and general-purpose protocols worth knowing about.

In summary, we are witnessing development of a new generation of Internet of Things-related products that will enhance our capabilities and our self-awareness. In this environment, Smartphones are likely to become the hub for personal-area networks, linking wearable gadgets such as healthcare monitors, smart watches, and display devices (e.g., Google Glass), as well as a variety of sensors embedded in clothes and shoes. This article has examined the design choices engineers have for generating and receiving data throughout this personal “electromagnetic bubble.” We have also examined a number of possible protocols and reviewed representative parts.

For more information on the parts mentioned in this article, use the links provided to access product information pages on the Hotenda website.